Jian Wang (王剑)

Jian Wang is a Ph.D. candidate (expected December 2025) at the College of Computing and Data Science (CCDS), Nanyang Technological University (NTU), Singapore, advised by Prof. Li Yi. His work focuses on code LLM security and intelligence.

Email  /  CV  /  Biography  /  Google Scholar  /  Twitter

Bio

Since 2019, Jian has been a research assistant at SMU and NTU. Previously, he was a researcher at Xiaomi AI Lab and a software engineer at 58.com.

He expects to receive his Ph.D. from NTU in 2025 and holds a B.A. in Software Engineering from Tianjin University (2011).

Jian’s research interests include machine learning algorithms and theory, adversarial methods for privacy-sensitive data, and their applications. He is particularly interested in security analysis, trend analysis, and privacy protection. Currently, he focuses on robust AI, developing high-fidelity adversarial attacks and defenses—such as trace-based repair for defense, adversarial motion-blur attacks, and facial-skew adversarial examples for deepfake detection.

Representative papers are highlighted.

Research

Automatic RNN Repair via Model-based Analysis
Xiaofei Xie, Wenbo Guo, Lei Ma, Wei Le, Jian Wang, Linjun Zhou, Yang Liu, Xinyu Xing
International Conference on Machine Learning (ICML), 2021
PDF / bibtex / Poster

We propose a lightweight model-based influence analysis to help understand and repair incorrect behaviors of an RNN. Specifically, we build an automaton to enable high-quality feature extraction and to characterize the stateful and statistical behaviors of an RNN over all training data.

Watch out! Motion is Blurring the Vision of Your Deep Neural Networks
Qing Guo, Felix Juefei-Xu, Xiaofei Xie, Lei Ma, Jian Wang, Bing Yu, Wei Feng, Yang Liu
Advances in Neural Information Processing Systems (NeurIPS), 2020
arXiv / bibtex

We introduce a motion-based adversarial blur attack (ABBA) that can generate visually natural motion-blurred adversarial examples.

FakeSpotter: A Simple yet Robust Baseline for Spotting AI-Synthesized Fake Faces
Run Wang, Felix Juefei-Xu, Lei Ma, Xiaofei Xie, Yihao Huang, Jian Wang, Yang Liu
International Joint Conference on Artificial Intelligence (IJCAI), 2020
arXiv / bibtex

Media coverage: Synced Review

Monitoring neuron behavior can help detect AI-synthesized fake faces, since layer-by-layer activation patterns may capture subtle features important for the detector.

Engineering

I forked this source code from jonbarron and xujuefei. Also consider Leonid Keselman’s Jekyll fork of this page.

ICP-1900352-1 link